Discussion
The M. arvalis populations are characterized by large-scale
genetic differentiation of Tshr , reflecting local adaptation to
annual temperature-photoperiod patterns, rather than latitude perse. Variation in Tshr sequence indicates that the M.
arvalis population can be subdivided into Eastern and Western European
clusters, indicating that they may belong to distinct genetic lineages
(Figs. 3, S3). This phylogeographical structure is consistent with that
found for mitochondrial cytochrome b gene sequences, and
microsatellite loci (representing nuclear DNA)(Haynes, Jaarola, &
Searle, 2003; Martínková et al., 2013; Stojak et al., 2016; Stojak,
Mcdevitt, Herman, Searle, & Wójcik, 2015), and justifies the analysis
of Western and Eastern European populations separately. The Western
versus Eastern divide could well be due to re-invasion of Northern
Europe from separate glacial refugia, and therefore separate
evolutionary events(Hewitt, 1999).
For insights into the geographical variation in Tshr , the
association of SNP frequencies with local climatic conditions was
examined. Here we showed that genetic variation in the vole Tshris better explained by local photoperiod-temperature patterns than by
latitude only. This may be caused by the temperature dependence of
vegetation growth. In house mice, genes (but not the Tshr ) that
show signals of selection are also associated with local average annual
ambient temperature, and are linked with clinal variation in phenotypic
aspects, such as body mass and metabolism(Ferris et al., 2021;
Phifer-Rixey et al., 2018). Interestingly, SNPs found in thyroid hormone
receptors, which are involved in regulation of seasonal reproduction in
the hypothalamus(Yoshimura et al., 2003), significantly correlated with
variation in average annual temperature(Ferris et al., 2021). This
suggests that genomic evolution of seasonal adaptation in house mice and
voles involves unique responses to genetic selection. Annual temperature
patterns not only depend on latitude, but also on longitude, altitude,
and other regional climatic variables like the Gulf Stream warming
European Atlantic coastal regions. Critical photoperiods in
pitcher-plant mosquitoes strongly correlated with altitude-corrected
latitude (r = 0.96), however, this measure does not integrate local
temperature patterns(W. E. Bradshaw et al., 2006; William E. Bradshaw,
1976; William E. Bradshaw & Lounibos, 1977). Deriving the regional
photoperiod-temperature ellipsoids may be better to account for such
regional climatic differences than latitude or altitude-corrected
latitude only. We post-hoc tested photoperiods at other temperature
thresholds, however this did not improve the results. Moreover, 6.6°C is
not an unreasonable temperature since grass growth is initiated at
5-10°C air temperature(Cooper, 1964; Peacock, 1975, 1976).
In addition, several SNPs correlated well with longitude and altitude
(Fig. 4G,J). Altitudinal gradients in seasonal timing of breeding have
been observed in deer mice (Peromyscus maniculatus borealis ),
with shorter breeding seasons at high elevations(Millar & Innes, 1985).
The pCPP at which a temperature threshold for grass growth initiation is
reached can be deduced from local photoperiod-temperature patterns, and
is here confirmed to be a strong determinant for distributional
variation in Tshr SNP frequency in Western European common vole
populations (Fig. 4M). Pairwise multilocus FST analysis
revealed that populations which differ in pCPP, also show greater
genetic distance in Tshr haplotypes (Fig. 3). These findings
indicate that seasonality is likely to be a selective force forTshr evolution in common voles, and imply that Tshr is an
important gene for genetic adaptation of the photoperiodic response
systems.
The observed genetic Tshr variation is unlikely to be caused by
isolation only, with the possible exception of the Orkney island
populations, which are geographically isolated from each other and from
mainland populations by the sea. Therefore isolation and genetic drift
may be a more important evolutionary force than natural selection in the
Orkney populations. Interestingly, the same SNPs appear to be related to
pCPP when the Orkney Island populations are excluded from the analysis.
This indicates that the results in Western Europe are not dominated by
the Orkney population’s data, and that the observed distribution ofTshr variation may be a sign of adaptive evolution likely
operating in response to photoperiod.
In Eastern European populations, none of the tested environmental
proxies are good predictors for Tshr SNP frequencies (Fig. 5).
These results indicate that the Tshr in the Eastern European
lineage is not linked to seasonal adaptation as observed in the Western
European lineage. Oceanic climates (Western Europe) are known for their
small annual temperature amplitudes, while continental climates (Eastern
Europe) are known for their large annual temperature amplitudes. These
climatic differences may have led to divergent evolutionary adaptation
of TSHR function, which may provide an explanation for the observed
longitudinal separation in genetic Tshr differentiation. Another
hypothesis is that photoperiodic genes other than the Tshr are
under selection for seasonal adaptation in Eastern European vole
populations.
SNPs associated with local pCPP were all synonymous or intronic
mutations. This suggests that these sites may be involved in regulatory
rather than structural variation. Five intronic SNPs were strongly
associated with pCPP in Western Europe (Fig. 4), of which two (i.e.
SNP-144 and -158) were strongly associated with altitude in Eastern
Europe (Fig. 5). Putative regulatory protein binding sites were
predicted for the intronic region, and revealed that intronic SNP-128,
which strongly correlates to pCPP (Fig. 4O), is located in a potential
SP1 (specificity protein 1) binding site(Höller, Westin, Jiricny, &
Schaffner, 1988; Ji, Casinghino, McCarthy, & Centrella, 1997).
Interestingly, SNPs closely located to this enhancer region, such as
SNP-158, are related to different environmental proxies in Eastern and
Western Europe (Fig. 4, 5, S2). It is tempting to speculate that
variation in and around this SP1 binding site sequence may influenceTshr transcription. Furthermore, there is strong evidence that
synonymous SNPs are not necessarily neutral as they can alter mRNA
expression, splicing, and structure, therefore having downstream effects
on protein expression(Chamary, Parmley, & Hurst, 2006; Hunt, Sauna,
Ambudkar, Gottesman, & Kimchi-Sarfaty, 2009). Synonymous polymorphisms
require different transfer RNAs (tRNA) to recruit the same amino acids
and may cause codon-bias. Synonymous tRNA vary strongly in frequency
between species and tissues (i.e. codon bias)(Dittmar, Goodenbour, &
Pan, 2006; Goodenbour & Pan, 2006). It is therefore possible that the
observed synonymous mutations in the TSHR may alter translation
efficiency within a species and tissue by changing the elongation
rate(Quax, Claassens, Soll, & Oost, 2015). Reduced elongation rate may
therefore result in lower protein abundance. Hence synonymous SNPs in
the Tshr gene could result in altered receptor abundance, changed
sensitivity to TSH and modified photoperiodic response. It is therefore
conceivable that synonymous SNPs in the Tshr gene are subject to
natural selection, and reflect local geographical adaptation. TSHR plays
a pivotal role in photoperiodic response in the pars tuberalis, but also
in thyroid hormone metabolism in the thyroid gland. Tissue-specific
functions of TSHR may benefit from genetic adaptation in photoperiodism
through synonymous SNPs, since tissue-specific tRNA expression, which
has been demonstrated in human and mouse tissues(Dittmar et al., 2006;
Pinkard, McFarland, Sweet, & Coller, 2020), may perhaps lead to altered
TSHR function in the pars tuberalis, but not in the thyroid gland.
Photoperiodic regulation of the reproductive system in deer mice has
been shown to vary with latitude, with weaker photoperiodic responses in
animals originating from lower latitudes(Dark et al., 1983). Moreover,
photoperiodic sensitivity in pitcher-plant mosquitoes correlated with
global warming, indicating the importance of season-length driving
evolution (genetic change) of photoperiodism during recent rapid climate
change(W. E. Bradshaw & Holzapfel, 2008; William E. Bradshaw &
Holzapfel, 2001a, 2006; William E. Bradshaw, Zani, & Holzapfel, 2004).
Our findings confirm that the Tshr gene is under selection, which
has previously been reported in chicken domestication in relation to
photoperiodic responsiveness(Karlsson et al., 2016; Rubin et al., 2010).
Future studies should determine whether the SNPs identified as seasonal
timing dependent genetic variation in the vole Tshr can indeed
alter, genetically based, photoperiodic responses. Such an approach will
confirm whether habitat-specific photoperiodic responses are indeed
regulated by means of functional TSHR adaptation. In vole populations
with later onsets of reproduction and shorter breeding seasons(Tkadlec,
2000), our results predict lower concentrations in the tanycytes ofTshr or lower TSH-binding affinities of Tshr haplotypes.
Optimal timing of reproduction, enhancing energetically demanding
pregnancy, and parental care, is necessary to maximize fitness in
temperate and northern seasonal environments. Tshr is an
essential gene in the pathway programming seasonal reproduction in
mammals. Herein, we show how onset of the favorable season over a wide
geographical range of the common vole, Microtus arvalis , explains
much of the genetic variation in the TSH binding site, hinge region, and
transmembrane domain of TSHR in Western but not Eastern Europe. Yet,
vole populations thrive in both regions. We therefore conclude that
different genetic mechanisms have been important in enabling vole
populations to exploit geographically distinct regions. Such
distinctions of how the genetic underpinnings of seasonal timing have
evolved over climatic gradients in nature will be important in
predicting how animals will adapt to new seasonal environments during
ongoing rapid climate change.